Strategic Planning
The 3 Levels of Agentic Shopping: Where Are We Now?
AI agents are starting to change how we shop, from discovery to delivery. But how far along are we, really? Drawing on Scot Wingo’s framework from Retailgentic, here’s a simplified take: three key stages in the rise of agentic commerce.
26 June 2025
5 min read
Level 1: Assisted Discovery
This is where most AI shopping experiences live today. Agents help you find products — showing product cards, summarising reviews, comparing prices. But they don’t buy for you. You still click out to a retailer and complete checkout yourself.
Examples: ChatGPT Shopping, Microsoft Copilot, Google Gemini.
Level 2: Agent-Guided Purchase
A step up. These agents not only help you decide what to buy — they start to take action. You can ask them to buy something, and they’ll complete the purchase (with limits). But they often depend on integrations like Amazon’s “Buy for Me” or Perplexity’s Pro. Payments are still clunky and usually brand-specific.
Examples: Amazon's “Buy for Me”, Perplexity’s “Buy with Pro”, early Google I/O demos.
Level 3: Autonomous Commerce
This is the endgame. AI agents that know your preferences, manage your budget, optimise loyalty points, and buy across retailers with a single wallet. They can re-order essentials, negotiate group discounts, and adapt in real time.
Examples: Not here yet — but the infrastructure is building fast. Payment APIs from Stripe, Visa, and startups like PayOS are laying the groundwork.
Why It Matters
For brands under £20m, this is a big shift:
Search is splitting: Voice, chat, and AI-native search are fragmenting the buyer journey.
Choice is collapsing: Agents narrow down thousands of options to a handful — if your product isn’t in the shortlist, it’s invisible.
Conversion moves upstream: The agent may “decide” before a shopper even lands on your site or Amazon page.
What Comes Next
We expect to see Level 2 agents gain traction this year, especially during Q4 holiday shopping. The jump to Level 3 will follow once multi-retailer payments and trust signals are in place.
If you sell on Amazon or DTC, now’s the time to rethink your content, visibility, and positioning for this shift. At LMO7, we’re tracking it closely.
This is where most AI shopping experiences live today. Agents help you find products — showing product cards, summarising reviews, comparing prices. But they don’t buy for you. You still click out to a retailer and complete checkout yourself.
Examples: ChatGPT Shopping, Microsoft Copilot, Google Gemini.
Level 2: Agent-Guided Purchase
A step up. These agents not only help you decide what to buy — they start to take action. You can ask them to buy something, and they’ll complete the purchase (with limits). But they often depend on integrations like Amazon’s “Buy for Me” or Perplexity’s Pro. Payments are still clunky and usually brand-specific.
Examples: Amazon's “Buy for Me”, Perplexity’s “Buy with Pro”, early Google I/O demos.
Level 3: Autonomous Commerce
This is the endgame. AI agents that know your preferences, manage your budget, optimise loyalty points, and buy across retailers with a single wallet. They can re-order essentials, negotiate group discounts, and adapt in real time.
Examples: Not here yet — but the infrastructure is building fast. Payment APIs from Stripe, Visa, and startups like PayOS are laying the groundwork.
Why It Matters
For brands under £20m, this is a big shift:
Search is splitting: Voice, chat, and AI-native search are fragmenting the buyer journey.
Choice is collapsing: Agents narrow down thousands of options to a handful — if your product isn’t in the shortlist, it’s invisible.
Conversion moves upstream: The agent may “decide” before a shopper even lands on your site or Amazon page.
What Comes Next
We expect to see Level 2 agents gain traction this year, especially during Q4 holiday shopping. The jump to Level 3 will follow once multi-retailer payments and trust signals are in place.
If you sell on Amazon or DTC, now’s the time to rethink your content, visibility, and positioning for this shift. At LMO7, we’re tracking it closely.